6 tweets · December 2020 · 6 likes · 0 retweets · read on Twitter
I've been musing... if there are 20M confirmed cumulative cases in the USA, wouldn't the number of not-confirmed cases be more like 100-200M? given how bad testing has been and the % of tests coming back positive being so high? Is anyone trying to answer this question?
I've been grumpy since the beginning that people keep talking about confirmed cases as if it's the territory, rather than a known-to-be-undercounted map. Basically another instance of the ubiquitous "no evidence" fallacy.
These guys back in April did some serology sample tests and found that there might be half a million true cases in LA County in California, versus the 8,000 confirmed active cases. dailynews.com/2020/04/20/l-a…
If only 10% of people infected are showing up in tests, this would imply that the USA is presently approaching herd immunity thresholds (*IF* people infected last spring are still immune) of >50%. Not saying this is so, but what's actually the best guess?
Let's do a poll! (I made it its own tweet so it could do its thing separate from this thread)
This post by @TheZvi cites a machine learning estimate of total cases that I'm not going to quote here next to the poll. I don't know to what extent it contains a detailed model or is just a basic number predictor. lesswrong.com/posts/CHtwDXy6…